Efficient Labeling of Vesicles with Lipophilic Fluorescent Dyes via the Salt-Change Method

Abstract

Fluorescent labeling allows for imaging and tracking of vesicles down to single-particle level. Among several options to introduce fluorescence, staining of lipid membranes with lipophilic dyes provides a straightforward approach without interfering with vesicle content. However, incorporating lipophilic molecules into vesicle membranes in an aqueous solution is generally not efficient because of their low water solubility. Here, we describe a simple, fast (<30 min), and highly effective procedure for fluorescent labeling of vesicles including natural extracellular vesicles. By adjusting the ionic strength of the staining buffer with NaCl, the aggregation status of DiI, a representative lipophilic tracer, can be controlled reversibly. Using cell-derived vesicles as a model system, we show that dispersion of DiI under low-salt condition improved its incorporation into vesicles by a factor of 290. In addition, increasing NaCl concentration after labeling induced free dye molecules to form aggregates, which can be filtered and thus effectively removed without ultracentrifugation. We consistently observed 6- to 85-fold increases in the labeled vesicle count across different types of dyes and vesicles. The method is expected to reduce the concern about off-target labeling resulting from the use of high concentrations of dyes.

Figure 1

Chemical structures of lipophilic fluorescent probes for vesicles. Lipophilicity (log P) values for octanol/water partition coefficient were calculated using a web tool (SwissADME) with a XLOGP3 model.

Figure 2

Fluorescent labeling of vesicles by salt-change method. (A, B) Schematics of direct staining (A) and salt-change (B) method for the labeling of cell-derived vesicles (CDVs) with DiI in an aqueous buffer. In the salt-change method, staining was performed in a low-salt buffer, then [NaCl] was increased to 150 mM to promote aggregation of DiI, and the aggregates were removed by filtration. (C) Schematic of single-vesicle imaging of DiI-labeled CDVs by TIRF microscopy. (D) Representative fluorescent images of DiI-labeled vesicles. Scale, 10 μm. (E) Numbers of DiI-labeled vesicles obtained by direct staining and salt-change method observed in TIRF images. Error bars, mean ± s.d. of n = 63 images. (F) Distributions of fluorescence intensity for the labeled vesicles. Inset, close-up view of the same curves; n = 56 (direct staining), n = 4671 (salt change) spots.

Figure 3

Applications of salt-change labeling. (A) Fluorescent labeling of mammalian EVs from NK-92 cells (NK-EV), bacterial outer-membrane vesicles from E. coli W3110 (E. coli OMV), and synthetic liposomes with comparison of the labeling methods. (B) Fold increase in labeling efficiency (vesicle count) calculated from (A). (C) Comparison of labeling methods for DiD and PKH67. For PKH67, results from a standard protocol (Supporting Information) is also shown. Error bars, mean ± s.d. of n = 20 (DiD) and 10 (PKH67) images. (D) Distributions of fluorescence intensity for the DiD- and PKH67-labeled vesicles shown in (C).

Figure 4

Size distribution and recovery of vesicles after salt-change labeling. (A) Nanoparticle tracking analysis (NTA) of vesicle size distribution for the unlabeled (gray) and 2 μM DiI-labeled (green) vesicles via the salt-change method. The size distributions (left panels) are shown with the corresponding total particle concentrations on right (bars). (B) NTA results for the salt-change labeling of NK-EVs with 0.2 μM DiI. In (A) and (B), error bars represent mean ± s.d. of n = 26–29 measurements. (C) Representative images of DiI-labeled EVs prepared by salt-change labeling; scale, 20 μm. (D) Number of DiI-labeled vesicles prepared with the indicated concentrations of DiI. Vesicle counts (left) from images such as shown in (C) are shown with the corresponding intensity distribution (right). Error bars, mean ± s.d. of n = 30 images.

Figure 5

Integrity of vesicle proteins after salt-change labeling (A) Schematic of single-vesicle pull-down and imaging of DiI-labeled NK-EVs containing PD-1-GFP. (B) Representative fluorescence images from the experiments described in (A). Scale, 5 μm. (C) Colocalization of DiI and GFP spots from the NK-EV images such as shown in (B). (D) GFP intensity distribution for the spots with DiI (labeled) and without DiI (unlabeled). (E) Schematic of NK-EV detection with PerCP-Cy5.5-conjugated PD-1 antibody. (F) Representative fluorescence images from the experiments described in (E). Insets show magnified views of the selected spots with and without DiI. Scale, 5 μm (on left) and 1 μm (insets). (G) Numbers of anti-PD-1 (PerCP-Cy5.5) spots with (yellow) and without (magenta) DiI signal. (H) Fraction of GFP spots detected by anti-PD-1 as a function of the presence of DiI.